Reducing Irreducible Background

and

Revealing the Unique Nature

of Neutrino Mass Using Fast Timing

Andrey ElaginUniversity of Chicago

University of Hawaii Physics Colloquium , March 13, 2017

Outline

• What can we learn about neutrinos by looking for neutrinoless double beta decay (0nbb-decay)?

• What instrumentation and experimental techniques are needed to find 0nbb-decay?

- Cherenkov/scintillation light separation

- development of the Large-Area Picosecond

Photo-Detectors (LAPPDTM)

2

3

Periodic Table of Elements

3

Helium Atom

Not to scale!

4

Periodic Table of Elementary Particles

(the Standard Model)The Higgs boson Example of a particle

“microscope”

As a graduate student I was searching for the Higgs

I now turned my attention to neutrinos

I’d like to build new kind of “microscopes” to study neutrinos 5

IIIIII

Bosons

Neutrinos

Ferm

ions

CDF

Discoveries and Instrumentation

State of the art instrumentation made these discoveries possibleFuture discoveries are waiting for new instrumentation

Nobel Prize 2013: the Higgs boson is foundP.Higgs and F.Englert

Nobel Prize 2015: neutrinos change while travel long distancesA.McDonald and T.Kojita

LHC

SNO Super-Kamiokanda

6

7

This Is What We Know

7

We Don't Know 95% of the Story

We have to build more instruments

More telescopes and “microscopes” are needed to

find out what are those 95%

Also we are not done with the ordinary matter yet! 8

Is the neutrino its own antiparticle?

It is possible because the neutrino has no electric charge

No other fermion can be its own antiparticle

It is not only possible, but may be necessary

- origin of matter-antimatter asymmetry in the universe

- why the neutrino mass is so tiny?

Neutrino Any other fermion

Search for neutrino-less double beta decay (0nbb-decay) is the most feasible way

to answer this question

A Question That Interests Me

9

Crisis in 1930

(known particles: g, p, e- )

beta decay: (A,Z) → (A,Z+1) + e- + ne

Meet the Neutrino

Electron energy spectrum

10

11

Letter by W. Pauli

11

12

”b-Strahlen”(A,Z) → (A,Z+1) + e- + n

e

4 particle interaction theory predicted

the electron energy spectrum remarkably well 12

Double Beta Decay

x

e-

e-

13

Double-Beta DisintegrationMaria Goeppert-Mayer

(A,Z) → (A,Z+2) + 2e- + 2ne

14

Double Beta Decay

Total energy of two electrons

15

Neutrinoless DecayIt is only possible if the neutrino is its own antiparticle

How can a particle be its own antiparticle?

16

Ettore Majorana

Noticed that symmetry of Dirac's theory allows to avoid

solutions with negative energies (antiparticles) for neutral

spin ½ particles

Fermi's theory of beta decay is unchanged if n = n

17

Giulio Racah

Proposed a “chain” reaction

(A,Z) → (A,Z+1) + e- +n

n + (A',Z') → (A',Z'+1) + e-

to distinguish between Dirac and Majorana neutrinos

18

Wendell Furry

Pessimistic conclusion about experimental prospects

to observe Racah's “chain” reaction:

- cross section is ~ 10-40

- no intense source for neutrinos (no reactors yet)

19

Wendell Furry

Proposed (A,Z) → (A,Z+2) + 2e- via virtual neutrino exchange

Quite optimistic experimentally:

0nbb-decay is a factor of 106 more favorable than 2nbb-

decay due to the phase factor advantage

V-A structure of week interactions is not known yet

20

Progress on Experimental Side1950 - Experimental limits on 0nbb exceeded predictions

(a hint that neutrino is a Dirac particle???)

1955 – R. Davis sets strong limits on n + 37Cl → 37Ar + e-

(interpreted as a proof that neutrino is a Dirac particle)

1957 – V-A nature of weak interactions → dramatic decrease in

probability of 0nbb-decay rate, also R.Davis' experiment doesn't solve

Dirac/Majorana questions for neutrinos

From reactor: n → p + e- +nR

At the target: nL + n → p + e- is allowed

nR + n → p + e- is forbidden by V-A couplings

helicity flip is required → 0nbb can't happen even for Majorana

neutrino if it has no mass

The fact that 0nbb-decay requires massive neutrino and lepton

number violation discouraged experimental searches21

22

Current StatusOscillation experiments established that neutrino is massive

and increased interest to 0nbb decay searches

Today we have many experiments

Why it has high priority?

KamLAND

SNO+

EXO

CUORE

MAJORANA

GERDA

Super-NEMO

In 2015 NSAC report 0nbb-decay was ranked as a high priority for US nuclear physics

22

Neutrinoless Decay Is UniqueIt may reveal the nature of neutrino mass

n → p + eL- + nR

nL + n → p + eL-

Even if neutrino is its own antiparticle nR

≠ nL

If neutrino is Majorana then nR is just a CP conjugate of nL , i.e. nLC = n

R

Therefore 0nbb-decay requires a mechanism for nLC n

Ltransition

Such transition is connected to a mass term in the LagrangianExample of a Majorana mass term: MNNCN 23

Possible extension of the SM Lagrangian

to introduce neutrino mass

See-Saw Mechanism

(νl , N R

c)(

0 mD

mD

TM RR

)(ν L

c

N R)

In the limit MRR >> mD the eigenvalues are

mD2/MRR (light neutrino)

MRR (heavy neutrino)vLvL

C

X

H

vLvLC

X

H

X

H

NR

“Effectively”

in the limit

MRR >> mD

This is exactly what's

needed for 0nbb-decay

eLeRX

HElectron mass term in the

Standard Model Lagrangian

meeLeR(Example of a Dirac mass term)

0nbb-decay provides access

to the neutrino mass mechanism 24

EXO (~200kg 136Xe)

KamLAND-Zen (~300 kg 136Xe,

before this Summer)

GERDA (~20 kg 76Ge)

Projections by

CUORE (~200kg 130Te)

SNO+ (0.8 ton 130Te)

SNO+ (8 ton 130Te)

S.M. Bilenky and C. Giunty Mod. Phys. Lett. A27, 1230015 (2012)

Experimental Sensitivity

T1/2-1 = G0n x |M0n|2 x m

bb2

Current best limit is setby KamLAND-Zen:T1/2 > 1.07x1026 yearsmbb < 61-165 meV

25

EXO (~200kg 136Xe)

KamLAND-Zen (~300 kg 136Xe,

before this Summer)

GERDA (~20 kg 76Ge)

Projections by

CUORE (~200kg 130Te)

SNO+ (0.8 ton 130Te)

SNO+ (8 ton 130Te)

S.M. Bilenky and C. Giunty Mod. Phys. Lett. A27, 1230015 (2012)

Experimental Sensitivity

T1/2-1 = G0n x |M0n|2 x m

bb2

Current best limit is setby KamLAND-Zen:T1/2 > 1.07x1026 yearsmbb < 61-165 meV

None of currently running or planned experiments

is sensitive to mbb

~1 meV 26

27

How to Find 0nbb-decay?

1) Choose an isotope

where 0nbb-decay is allowed

2) Wait for emission of

two electrons with the

right total energy

Isotopes

Q-value(Total energy of 2 electrons),

MeV

Natural abundance,

%

27

Challenge #1

Life-time for 0nbb-decay is more than > 1026 years

This is much longer than the age of the universe

Solution: look at many atoms at the same time

- Avogadro number is large NA = 6x1023

- one ton of material can have >1027 atoms

- even with one ton we are talking about ~10 events per year

Very Small Decay Probability

28

Challenge #2

Solution: good energy resolution

2nbb

0nbb

Background from 2nbb-decay

29

Challenge #3

Solution: purification and shielding

These decays are a factor of ~1016 more likely than 0nbb-decay

There are 3g U-238 and 9g of Th-232 per ton of rockNatural Radioactivity

30

Ideal 0nbb-decay Experiment

1) Large mass (more nuclei at the same time)

2) Good energy resolution (discriminate from 2nbb-decay)

3) Purification and shielding (natural radioactivity)

T1/2 ~

31

New Challenge for a Large Detector

8B solar neutrino interactions become dominant background

This is irreducible background without event

topology reconstruction

Electron scattering of neutrinoscoming from 8B-decays in the sun

n

n

e- e-

32

The largest background is coming from 8B solar neutrinos

It has only 1 electron, while nbb-decay has 2 electrons

Is it possible to separate two-track and one-track events

using Cherenkov light in a liquid scintillator detector?

Background Budget at SNO+

33

Can We See This?

R=6.5m

Simulation of a back-to-back 0nbb event

2014 JINST 9 P06012

34

Double-Beta Decay Kinematics

• Distinct two-track topology with preference to be “back-to-back” • Electrons are above Cherenkov threshold

Angle (cosQ) between two electrons Kinetic energy of each electron

Cherenkov threshold

Event generator based onphase factors from J.KotilaPRC 85 (2012) 034316

35

Can We Detect Cherenkov Light?

Scintillation emission is slower

Longer wavelengths travel faster

Cherenkov light arrives earlier

Scintillation light is more intense and

Cherenkov light is usually lost in liquid

scintillator detectors e-

370 nm 0.191 m/ns600 nm 0.203 m/ns

~2 ns difference over 6.5m distance

Scintillation model based on KamLAND-Zen simulation

36

Can We Detect Cherenkov Light?PE arrival times, TTS=100 ps

• Cherenkov light arrives earlier• Need good timing to see the effect

37

Directionality and Vertex Reconstruction

5 MeV

2.1 MeV

1.4 MeV

Directionality VertexSimulation:- single electrons along X-axis

at the center of 6.5m sphere - KamLAND scintillatorReconstruction:WCSim adapted for low energy

2014 JINST 9 P06012

Directionality “survives” some detector effects

Vertex resolution is promising

Directionality is already a handle on 8B eventsSolar neutrinos come from the sun and outgoing electrons “remember” that 38

Directionality or Topology?Idealized event displays: no multiple scattering of electrons, all PEs, QE=30%

S0

S1

S2

S3

Rotation invariant power spectrum

Spherical harmonics analysis

0nbb-decay 8B eventCherenkov PEs

Scintillation PEs

S power spectrum

39

Early Light Topology

Cherenkov PEs

Scintillation PEs

S0

S1

S2 S

3

Why spherical harmonics?• Spherical harmonics analysis is a natural and

“easy” choice for a spherical detector• Advanced machine learning techniques will

do even better• Understanding of requirements on hardware

components is now a much higher priority – thoseare hard to change once the detector is built

Early PE: 0nbb-decay Early PE: 8B event

S power spectrum

Realistic event displays: early PEs only, KamLAND PMTs QE: Che~12%, Sci~23%

40

0nbb vs 8B

Multipole moment l=0 Multipole moment l=1

Simulation details: 6.5m radius detector, scintillator model from KamLAND simulation

TTS=100 ps, 100% area coverage, QE(che) ~12, QE(sci) ~23%

41

0nbb vs 8B

Ideal vertex, central events onlyScintillation rise time 1 ns

Key parameters determining separation of 0nbb-decay from 8B Scintillator properties (narrow spectrum, slow rise time)

Photo-detector properties (fast, large-area, high QE, red-sensitive)

42

0nbb vs 8B

Vertex res 5cm, events within R<3mScintillation rise time 1 ns

Vertex res 5cm, events within R<3mScintillation rise time 5 ns

Background rejection factor = 2 @ 70% signal efficiency

Background rejection factor = 3 @ 70% signal efficiency

For details see NIM A849 (2017) 102

Other backgrounds (gammas, alphas, 10C, etc) also have distinct topologiesEvent reconstruction in liquid scintillator would enable new opportunities

43

Illustration from a presentationby Gabriel Orebi Gann

THEIA

Plot credit: Andy Mastbaum

Broad detector R&D program to realize THEIA

• 50kt detector• 50% reduction of 8B• 0.5% natTe loading • 50t 130Te after fiducial cuts• 15 meV after 10 years

Multipurpose detector(including neutrino oscillation physics)

Potential for 0nbb-decay search

44

NuDot - Directional Liquid Scintillator

• 140 2” fast PMTs for timing• 72 10” regular PMTs for energy resolution

R&D Towards Large Scale Detector

emissionabsorption

2.2 m

• Nanocrystals of CdS, CdSe, CdTe• Interesting optical properties• nbb-decay candidates• Q-dots can be suspended in organic

solvents and water• In-depth R&D is needed to evaluate

Q-dots potential

Q-dots

46

Under construction at MIT, led by L. Winslow

Goals• Demonstrate directionality and event

topology reconstruction using che/sciseparation by fast timing- ideally by measuring 2nbb-decay

• Study scintillators, including quantum dots

OTPC installed at MCenter, FNAL

Eric Oberla PhD thesis

Single event

0 ns

20 ns-570 mm -160 mm

Example event

Optical Tracking Demonstration180-channel PSEC4 system

Typical event(thru-going μ)

NIM A814 (2016) 19

47

47

The ANNIE Experiment• Measure neutron multiplicity in neutrino-nucleus interactions• R&D towards water-based neutrino detection technology• Explore optical tracking using novel photo-detectors

48

ANNIE installation at Fermilab

Data taking is ongoing

PMT by HamamatsuLarge area, but slow…

photo credit: http://kamland.stanford.edu

photo credit: E.Oberla PhD thesis

MCP-PMT by Photonis:Fast, but small…

Photo-Detector Options

5 cm

17-20”

49

Photo-Detectors

Photo-Multiplier Tube (PMT) is a classical example of a photo-detector

- use photo-electric effect to convert a photon to an electron

- use secondary electron emission (SEE) to amplify the signal

Uncertainty on the electron path causes uncertainty on the signal timing

The shorter the electron path the better the time resolution

No existing fast photo-detectors can cover large area at a reasonable cost50

LAPPDTM

Micro-Capillary Arrays by Incom Inc.

Material: borofloat glass Area: 8x8” Thickness: 1.2mm Pore size: 20 mm

Open area: 60-80%

Atomic Layer Deposition (ALD)- J.Elam and A.Mane at Argonne

(process is now licensed to Incom Inc.)- Arradiance Inc. (independently)

Micro-Channel Plates (MCPs)

20x20 cm2

~15mm

Large-Area Picosecond Photo-Detector

Single PE time resolution <50ps

51

LAPPD Prototype Testing ResultsSingle PE resolution

RSI 84, 061301 (2013), NIMA 732, (2013) 392

NIMA 795, (2015) 1See arXiv:1603.01843

for a complete LAPPD bibliography

Demonstrated characteristics:single PE timing ~50psmulti PE timing ~35 ps

differential timing ~5 psposition resolution < 1 mm

gain >107

Reconstruction of the laser beam footprint

52

LAPPDTM Commercialization

Slide courtesy of Incom Inc. 53

Goal of the R&D Effort at UChicago

Affordable large-area many-pixel photo-detector systems

with picosecond time resolution

LAPPD module 20x20 cm2 Example of a Super Module

UChicago goal is to enable high volume production at Incom so that LAPPDTM become available for HEP community

• High volume production can be challenging for vacuum transfer process• We are exploring if a non-vacuum transfer process can be inexpensive

and easier to scale for a very high volume production

54Production rate of 50 LAPPDs/week would cover 100 m2 in one year

In-Situ LAPPD Fabrication

UChicago PSEC Lab

Simplify the assembly process by avoiding vacuum transfer:make photo-cathode after the top seal

(PMT-like batch production)

Heat only the tilenot the vacuum vessel

Intended for parallelization

55

In-Situ Assembly Facility UChicago

Looking forward towards transferring the in-situ process to industry

The idea is to achieve volume production by operating many small-sizevacuum processing chambers at the same time

UChicago PSEC Lab

56

First Signals from an In-Situ LAPPD

Near side: reflection from unterminated far end

Far side: reflection is superimposed on prompt

Source

far sidenear side

Source

Readout(50-Ohm transmission line)

(Sb cathode)

Readout(50-Ohm transmission line)

The tile is accessible for QC before photo-cathode shot This is helpful for the production yield

April, 2016

57

First Sealed In-Situ LAPPDAugust 18, 2016

Flame seal by J.Gregar, Argonne

58

(Cs3Sb photo-cathode)

Gen-II LAPPD

10 nm NiCr ground layer insideis capacitively coupled

to an outside 50 Ohm RF anode

NiCr-Cu electrodingfor the top seal

Ground pins

Two tubulation portsfor the in-situ PC synthesis

(improved gas flow)

Monolithic ceramic body

• Robust ceramic body• Anode is not a part of

the vacuum package• Enables fabrication

of a generic tile fordifferent applications

• Compatible with in-situ and vacuum transfer assembly processes

Joint effort with Incom Inc. via DOE SBIR 59

January, 2017

We need lots of stuff and we often build what we need

Lots of Hands On Experience

This is fun!

60

Dirac/Majorana nature of the neutrino is a fundamental question

Search for 0nbb-decay is the most feasible approach to answer this question

Very large detector mass (kilo-ton) is required to probe small mbb

8B solar neutrinos become dominant background - traditionally viewed as

irreducible

Directionality and event topology provide handles on 8B background

Detector R&D is ongoing to demonstrate event topology reconstruction

using Cherenkov/scintillation light separation

Fast timing is critical and there has been lots of progress in

the development of LAPPDTM

61

Take Away Messages

Thank You

62

62

Only Three Flavors*

Nn

= 2.9840+- 0.0082

63

Neutrino Mixing

Flavor eigen states(interaction)

Mass eigen states(propagation)

64

Neutrino Mass Hierarchy

65

66

0nbb vs 2nbbEvents within 5% of the end point

Event generator from L.Winslow based on phase factors from PRC 85, 034316 (2012)

by J. Kotila and F. Iachello

My e-mail exchange with Jenni Kotila:

“…The angular correlation is basically the a^(1)/a^(0), where a^(i) are defined in Eq. (24)

for 2nbb and in Eq. (51) for 0nbb. In case of 0nbb only thing that matters are the electron wavefunctions but in case of 2nbb there are these additional factors that are a combination of <K_N> and <L_N>, that are defined in Eq. (23) and include the electron energies, the

neutrino energies and the closure energy. So even with small neutrino energies, for example e_1=0.749Q, e_2=0.249Q, w_1=0.002Q, w_2=0 a factor of 0.4329 is obtained. Regarding the question about the situation for different isotopes, the closure energy entering the equations is

different for each isotope and can be approximated by 1.12A^(1/2) MeV…”

67

Directionality of Early Photons

C.Aberle, A.Elagin, H.Frisch,

M.Wetstein, L.Winslow

2014 JINST 9 P06012

68

½ Q (116Cd) =1.4 MeV ½ Q (48Ca) =2.1 MeV

Light yield: Cherenkov vs scintillation

What About Lower Energies?

0nbb vs 8B

Vertex res 5cm, events within R<3mSci rise time 1 ns

18

Ioverlap = 0.79

0nbb vs 8B

Vertex res 5cm, events within R<3mSci rise time 5 ns

19

Ioverlap = 0.64

71

Off-Center Events

72

0nbb-decay vs 10Ctwo-track vs a “complicated” topology

10C decay chain:

• 10C final state consist of a positron and gamma(e+ also gives 2x0.511MeV gammas after loosing energy to scintillation)

• Positron has lower kinetic energy than 0nbb electrons• Positron scintillates over shorter distance from primary vertex• Gammas can travel far from the primary vertex

10C vs 0nbb-decay: photons arrival time profile

Diagram by Jon Ouellet

10C background can be large at a shallow detector depth

TTS=100 ps

73

0nbb-decay vs 10CPhotons count in early light sample

Time profile for events uniformlydistributed within the fiducial volume, R<3m

Vertex resolution of 3cm is assumed

Spherical harmonics help here too

Disclaimer: there are other handles on 10C that are already in use (e.g., muon tag, secondary vertices). Actual improvement in separation power may vary.

TTS=100 ps

74

75

76

77

Production rate of 50 LAPPDs/week

would substitute all PMTs at SNO+

in 3-4 years

How many LAPPDs are needed?• NuDot needs up to 72 LAPPDs (small-scale prototype with a path to a very large

directional liquid scintillator detector for 0nbb-decay)

• ANNIE needs 20-100 LAPPDs (water Cherenkov detector at Fermilab)

• KamLAND-Zen and SNO+ may benefit from LAPPDs but would need thousands of LAPPDs

• THEIA would need over 20,000 LAPPDs for just a 10% photo-coverage

Need for High Volume ProductionKey applications

• Cherenkov/scintillation light separation to reconstruct 0nbb-decay event topology

• Optical tracking

• Particle identification by time-of-flight (colliders and fixed-target experiments)

• Medical imaging, proton therapy, nonproliferation, quantum imaging

4

Early Adopters of LAPPD

Some examples of early adopters:

• ANNIE – Accelerator Neutrino Neutron Interactions Experiment• Cherenkov/Scintillation light separation for particle ID• Optical Time Projection Chamber• TOF measurements at Fermilab Test Beam• There are many more (lots of interest shown at the “Early Adopters

Meeting” hosted by Incom Inc. in 2013)

9

Putting first LAPPD tiles into real experimental settings for testingis the highest priority

FlatDot Demonstration

15 cm Quartz Vial

• Intermediate step towards 1m3 spherical NuDot- e.g. detection of Cherenkov “rings” from low energy

electrons using a tagged Compton source• Testing different scintillator cocktails• Readout testing

2” PMTs with TTS=300ps

22

Note: there is an independent effort on Che/Sci light separation - the CHESS experiment at Berkeley by G. Orebi Gann et al., aXiv:1610.02011 and 1610.02029

Time (ns)

Raw pulses (the top two channels are the trigger)

Event display after corrections

In-Situ Assembly StrategySimplify the assembly process by avoiding vacuum transfer:

make photo-cathode after the top seal(PMT-like batch production)

Step 1: pre-deposit Sb on the top window prior to assemblyStep 2: pre-assemble MCP stack in the tile-baseStep 3: do top seal and bake in the same heat cycle

using dual vacuum systemStep 4: bring alkali vapors inside the tile to make photo-cathodeStep 5: flame seal the glass tube or crimp the copper tube

UChicago processing chamber

33

Heat only the tilenot the vacuum vessel

Intended for parallelization

Sb layer only Cs-Sb photo-cathode

In-Situ Photo-Cathode

Cs

Relative QE measurementFirst in-situ commissioning run (Summer 2016)- saw the first photo-current response

from in-situ photo-cathode- measured relative QE (absolute QE is tricky

due to DC current through the whole stack)- demonstrated a sealed tile configuration

- no QE drop for 2 weeks after the valve to the pump was closed

- no QE drop for 3 weeks after flame seal

Note on this commissioning run:PC is very thick for transmission mode operation

(initial 20nm of Sb translates into ~80nm of Cs-Sb) 38

near center

far

July, 2016

Gen-II LAPPD: “inside-out” anode

Custom anode is outside

Compatible with high rate applications

For details see arXiv:1610.01434(submitted to NIM)

Choose your own readout pattern

42

Inside-out Anode Testing

Evan Angelicoand

Todd Seiss

43

arXiv:1610.01434

85

LAPPD Electronics @ UChicago

Delay-line anode

- 1.6 GHz bandwidth

- number of channels

scales linearly with area

PSEC-4 ASIC chip

- 6-channel, 1.5 GHz, 10-15 GS/s

30-Channel ACDC Card (5 PSEC-4) Central Card

(4-ACDC;120ch) 4

NIM 711 (2013) 124

NIM 735 (2014) 452

Can you make PC after Sb was exposed to air?

Luca Cultrera at Cornell

What about noise in the MCPs after Cs-ation?

Matt Wetstein

Indium seal recipes exist for a long time

PLANACONTM

(MCP-PMT by Photonis)

5 cm

Make larger photo-detectors

Our recipe scales well to large perimeter

Simplify the assembly process

Our recipe is compatible with PMT-like batch

production

Why do we need another indium seal recipe?

We adapted NiCr-Cu scheme

from O.Siegmund at SSL UC Berkeley

In-Situ Process Pre-requisiteReliable hermetic seal over a 90-cm long perimeter

Indium Solder Flat Seal RecipeInput:

• Two glass parts with flat contact surfaces

Process:

• Coat 200 nm of NiCr and 200 nm of Cu

on each contact surface (adapted from

seals by O.Siegmund at SSL UC Berkeley)

• Make a sandwich with indium wire

• Bake in vacuum at 250-300C for 24hrs

Key features:

• A good compression over the entire perimeter

is needed to compensate for non-flatness and

to ensure a good contact

• In good seals indium penetrates through entire

NiCr layer (Cu always “dissolves”)

glass window(8.66x8.66”)

glass frame(sidewall)

Sealed LAPPD tile

This recipe is now understood

It works well over large perimeters35Metallization and compression are critical

Metallurgy of the SealModerate temperatures and short exposure time:

• A thin layer of copper quickly dissolves in molten indium

• Indium diffuses into the NiCr layer

Depth profile XPS

The ion etch number is a measure for the depth of each XPS run

Layer depth (uncalibrated)

XPS access courtesy ofJ. Kurley and A. Filatov at UChicago

Glass with NiCr-Cu metallization exposed to InBi at ~100C for <1hrs

(it seals at these conditions)

InBi was scraped when still above melting (72C)

Low melting InBi alloy allows to explore temperaturesbelow melting of pure In (157C)

Metallurgy of the Seal

SEM and EDAX of the metal surfacescraped at the interface

SEM/EDAX data courtesy of J. Elam at Argonne

Glass with NiCr-Cu metallization bonded by pure In at ~250C for 2hrs

(it seals at these conditions)

Cut and scrape at the metal-glass interface

In:77-86%

Ni: 4-12%

Cu: 1-6%

High temperatures and long exposure time

• Indium penetrates through entire NiCr layer

Cr: 1-16%

glass window

sidewall

indium seal

Metallurgy of a Good SealHigher temperatures and longer exposure time

• Indium penetrates through entire NiCr layer

XPS of the glass side of the interface

XPS data courtesy of A. Filatov at UChicago

Cut and scrape at the metal-glass interface

Inte

ns

ity (

a.u

.)

900 800 700 600

Ni(2p)Cr(2p)

control surface

scraped region

In(3p)

Binding energy, eV

Glass with NiCr-Cu metallization bonded by pure In at ~350C for 24hrs

(it seals at these conditions)

We now reliably seal at 250-300C for 12-24hrs

93

Slide credit: Henry Frisch

Slide courtesy of R. Darmapalan and R. Wagner

Argonne 6x6 cm2 Photo-Detectors

8